-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Anders Boström wrote: >>>>>> "BM" == Bill Moran <[EMAIL PROTECTED]> writes: > > BM> In response to Ryan Novosielski <[EMAIL PROTECTED]>: > > >> > - A switch (mostly 3Com switches in my experience) that run in > half-duplex > >> > mode, which slows network traffic down by about a factor of 10. > >> > >> Cisco does this just as often, if not more often. A little surprising to > >> find that the top 2 can't seem to compete on the same level as a D-Link > >> switch from Radio Shack. ;) > > BM> If you read Cisco's docs, they make the claim that these problems are > BM> per-spec. > > BM> My understanding of the argument is that if you manually set the speed > BM> and duplex, you have disabled auto-negotiation. If the other end > BM> tries to auto-negotiate, it will be able to detect the speed, just by > BM> dumb luck of how the protocol works, but it will _consistently_ mis- > BM> detect the duplex, again because of dumb luck of the protocol > > BM> Their argument seems to be that this behaviour is per the specs. If > BM> a D-Link does it differently, then D-Link is doing it "wrong", even if > BM> it's doing it more intuitively. > > BM> If Cisco is correct, then it would seem as if the spec were written > BM> poorly. > > This is a bit off-topic, but anyway... > > The standard spec is IEEE 802.3 std 200[025], and it specifies the > auto-negotiate (auto-neg) protocol (among other things). If both ends > implements auto-neg as specified, everything works fine and you get > the best link-settings both supports. If you turn off auto-neg, you > have to set the port to the same setting in both ends. The only reason > to turn off auto-neg is that one end isn't implementing auto-neg > according to spec (and I'd say it is broken). In that case you have to > setup the ports in both ends manually in order to guarantee that it > works. > > If one end is set to auto-neg and the other is hard set to 100 Mbps > full duplex with auto-neg turned off, the auto-neg port will end up in > 100 Mbps half duplex in most cases. > > The reason for this is that it will not receive any answer to the > auto-neg messages, and will then in most implementations resort to > parallel detection. Parallel detection can sense the speed (10 or 100 > Mbps) but it is fundamentally impossible to know if the other end is > set in full or half duplex, without auto-neg. > > All bridges and stations supporting full duplex are supposed to > implement the standard, and use the auto-neg protocol. A peer-partner > not implementing auto-neg is assumed to only support half > duplex. That's why the auto-neg end are set in half duplex. > > Why would you turn off auto-neg if you want automatic > link-configuration???
Because what I want and what I get are not the same thing. I'd consider running a more specific test next time I get the opportunity with a new piece of hardware, but it seems as if with our current equipment (assuming I trust our network people... which... yeah, I guess that is questionable at times), I leave my end alone (autonegotiate by default) and have the switch port left alone (autonegotiate by default), I get 100/HD as my link type. - -- ---- _ _ _ _ ___ _ _ _ |Y#| | | |\/| | \ |\ | | |Ryan Novosielski - Systems Programmer III |$&| |__| | | |__/ | \| _| |[EMAIL PROTECTED] - 973/972.0922 (2-0922) \__/ Univ. of Med. and Dent.|IST/AST - NJMS Medical Science Bldg - C630 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.5 (MingW32) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFFLk92mb+gadEcsb4RAlbFAJ9yLGgr37ZR5Sata4IIloO/ucZcEgCbB9IX 6AXGKxDSvXx5B5E9edtcbWs= =p9N/ -----END PGP SIGNATURE----- ------------------------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 _______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
